Low-Energy resonances in the 22 Ne(p,γ) 23 Na reaction directly observed at LUNA

. The neon-sodium cycle of hydrogen burning inﬂuences the synthesis of the elements between 20 Ne and 27 Al in AGB stars and classical novae explosions. The 22 Ne(p, γ ) 23 Na reaction rate is very uncertain because of a large number of unobserved resonances lying in the Gamow window. A new direct study of the 22 Ne(p, γ ) 23 Na reaction has been performed at the Laboratory for Underground Nuclear Astrophysics (LUNA) using a windowless gas target and two HPGe detectors. Several resonances have been observed for the ﬁrst time in a direct experiment.


Introduction
The neon-sodium cycle of hydrogen burning contributes to the synthesis of the isotopes between 20 Ne and 23 Na in Red Giant Branch (RGB) stars, Asymptotic Giant Branch (AGB) stars and classical novae explosions.The synthesis of sodium in RGB stars is still puzzling.Observations of galactic globular clusters show that the surface abundance of sodium in RGB stars anticorrelates with the oxygen abundance [1].A possible explanation for this anticorrelation involves the pollution of the interstellar medium with material processed through hydrogen burning reactions at high temperatures in AGB stars.In the hydrogen burning shell of AGB stars, oxygen is efficiently destroyed by the CNO cycle and sodium is mainly produced by the 22 Ne(p,γ) 23 Na reaction of the NeNa cycle [2].Another scenario where the 22 Ne(p,γ) 23 Na reaction is active are classical novae explosions.A sensitivity study showed that the 22 Ne(p,γ) 23 Na reaction rate uncertainty strongly affects the final abundances of neon, sodium and magnesium isotopes, calling for new experimental efforts on the 22 Ne(p,γ) 23 Na cross section [3].The 22 Ne(p,γ) 23 Na Gamow window for AGB stars and classical novae extends from 50 to 600 keV.In this energy range, the proton capture on 22 Ne is dominated by a large number of resonances.None of the resonances below 436 keV has ever been observed in either direct or indirect experiments (tab.1).Moreover, the mere existence of the three resonances at 71, 105 and 215 keV is still debated since it has been tentatively reported in [4] but has not been observed in subsequent experiments [5,6].As a consequence, the 22 Ne(p,γ) 23 Na reaction rates reported in the widely adopted compilations by NACRE [7] and C. Iliadis et al. [8] are up to three orders of magnitude discrepant in the energy range of interest for hydrogen burning in AGB stars and novae.

Experimental setup
The Laboratory for Underground Nuclear Astrophysics (LUNA) is located at Gran Sasso National Laboratories, Italy, where the 1400 meters of rocks dominating the laboratory guarantee a reduction of six orders of magnitude in the cosmic muon flux and a reduction of three orders of magnitude in the neutron flux.
The 400 kV electrostatic accelerator provides high intensity (∼200 µA) proton or alpha beam.The beam can be delivered either to a solid or to a gas target.Different gamma-ray or particle detectors can be used, depending on the nuclear reaction to be studied [10].
As a first step, a feasibility test was performed using the setup of the previous 2 H(α,γ) 6 Li experiment [11].For this test, neon gas with natural isotopic composition was used (90.48% 20 Ne, 0.27% 21 Ne and 9.25% 22 Ne).The aim of the test was to study the possible sources of beam induced background, and to have some hints on the sensitivity to the 22 Ne+p resonant cross section.
During the test, runs on the 22 Ne(p,γ) 23 Na resonances at 105, 158, 186 and 259 keV have been performed.Despite the use of non enriched gas (only 9.25% 22 Ne) and a setup which was not optimized for this measurement, gamma-rays from the 186 keV resonance have been observed in a 12 hours run.This resonance was never observed in previous experiments, and the literature upper limit for the resonance strength is ωγ < 2.6 • 10 −6 eV.Thanks to the LUNA observation it was possible to provide a lower limit ωγ ≥ 0.12 • 10 −6 eV for the 186 keV resonance strength and thus to improve the literature information on this resonance [12].Following the feasibility test, the characterization of the setup for the first experimental campaign on the 22 Ne(p,γ) 23 Na resonances was started.In this phase the gas density profile along the beam path was first measured without beam using natural neon gas and a dedicated target chamber.Then the beam heating effect in natural neon gas has been measured for the first time with the resonance scan technique, using the intense and well known 21 Ne(p,γ) 22 Na resonance at 271.6 keV beam energy [13] and a collimated NaI detector [12].For the 22 Ne(p,γ) 23 Na cross section measurement, a proton beam was delivered to the windowless gas target filled with 99.9% enriched 22 Ne.The gamma-rays emitted in the de-excitation of 23 Na were detected by two HPGe detectors collimated at 55 • and 90 • with respect to the beam direction.
In order to reduce the environmental background at gamma ray energies below 3 MeV, the two detectors were surrounded by a 4 cm thick copper shielding and a 25 cm thick lead shielding.

Results
During about five months of data taking, all the resonances between 70 and 334 keV have been investigated.The resonances at 158, 186 and 259 keV have been observed for the first time in a direct experiment.For these resonances, the complete excitation function was measured and then a long run at the energy of maximum yield was performed.New gamma decay modes have also been observed for the three resonances detected.
For the non-detected resonances new upper limits have been measured.The new upper limits are two to three orders of magnitude lower than the previous direct measurement, proving the improvement in sensitivity that can be achieved in underground experiments.Fig. 1 shows the updated reaction rate for the 22 Ne(p,γ) 23 Na reaction, calculated including the new LUNA results.Literature reaction rates adopted in references [7] and [8] are also shown for comparison.Thermonuclear reaction rate from LUNA, NACRE [7] and Iliadis et al. [8] normalized to [8].Colored bands represent the reaction rate uncertainties.
The final results will be presented and discussed in detail in a forthcoming publication [15].